TRUMP SIGNS ORDER GIVING GOVERNMENT EARLY EYES ON AI MODELS BEFORE RELEASE — COMPANIES CAN VOLUNTEER OR FACE THE ALTERNATIVE President Trump signed an executive order on June 2nd directing federal agencies to build a framework allowing AI developers to voluntarily submit frontier models to the government for up to 30 days before public release. The National Security Agency will lead a classified benchmarking process to determine which models qualify as covered frontier models and therefore fall under the program’s scope. The Treasury Department, NSA, and CISA are jointly responsible for designing the framework within 60 days. The order is structured as voluntary, which is the key word. No company is currently compelled to hand over its models. But the government is establishing the apparatus to review them, and the order also directs agencies to prioritize criminal enforcement against AI-enabled cybercrimes, which is the stick underneath the voluntary carrot. The order is framed around national security rather than consumer safety, which marks a deliberate contrast to the Biden administration’s AI executive order from 2023, which leaned heavily on safety testing language. The Trump White House is betting that leading on AI capabilities while managing the most acute security risks is a better strategy than broad regulation. The AI industry has largely praised the approach, preferring government engagement over European-style oversight. Whether the voluntary framework holds as models get more powerful is the open question nobody can answer yet. Keywords: Trump AI executive order, frontier AI models cybersecurity, NSA AI review, AI government oversight, AI national security policy